A probabilistic approach towards source level inquiries for forensic soil examination based on mineral counts. (November 2021)

• Record Type: Journal Article
• Authors: Lim, Yu Chen; Marolf, André; Estoppey, Nicolas; Massonnet, Geneviève
• Is Part Of: Forensic science international. Volume 328(2021)
• Publication Date: 2021-11
• Subjects: Forensic geology -- Bayesian approach -- Heavy mineral fraction -- Light mineral fraction -- Polarized light microscopy -- Scanning electron microscopy -- Multivariate statistics -- Inter-variability -- Intra-variability -- Blind tests; Medical jurisprudence -- Periodicals; Chemistry, Forensic -- Periodicals; Forensic Medicine -- Periodicals; Médecine légale -- Périodiques; Chimie légale -- Périodiques; Gerechtelijke geneeskunde; Gerechtelijke chemie; Gerechtelijke psychiatrie; Chemistry, Forensic; Medical jurisprudence; Electronic journals; Periodicals; Electronic journals; 614.1
• DOI: 10.1016/j.forsciint.2021.111035
• Languages: English
• ISSNs: 0379-0738

Abstract

Highlights: Light mineral fraction was inadequate in differentiating between sources of soil samples. Heavy mineral fraction had several minerals whose proportions allow for discrimination by location. Statistical methods highlight these differences and permit the building of a classification model. Probabilistic approach allows for presentation of results in an informative manner to answer source level questions for comparison of soil samples. Abstract: Forensic soil examination has a well-established foundation in forensic science, this is in part due to the widely varied and complex nature of soil. Within this domain, mineral suite studies are a commonly utilized tool in soil examination. However, statistical or probabilistic approaches towards the interpretation of results from such analysis are lacking and this study aims to fill that gap. Soil samples from four different locations in the city of Lausanne, Switzerland were sampled and their mineral fractions, light and heavy of size between 90 and 180 µm, were studied utilizing microscopical methods. First, the light minerals were identified and counted by employing scanning electron microscopy coupled with energy-dispersive X-ray spectroscopy (SEM-EDS). Second, the heavy minerals were identified and counted manually under a polarized light microscope (PLM). The resulting count data were subjected to various multivariate statistical treatments such as principal components analysis (PCA), hierarchical clustering analysis (HCA), and linear discriminant analysis (LDA). These methods assist in identifying pertinent variables and subsequently in building various classification models. The validities of these models were then tested and evaluated using blind tests. Finally, these methods demonstrate how a probabilistic approach can be taken in the interpretation of the results to answer source level questions.